In R01AG17761, 150 patients with mild cognitive impairment (MCI) and 63 age and sex-matched healthy control subjects are followed. With an average of 5 years'follow-up, specific neuropsychological test scores, odor identification deficits, patient-informant discrepancy in report of function, MRI hippocampal and entorhinal atrophy are highly significant predictors of conversion to AD (41 converters to date), apolipoprotein E e4 genotype is significant only in older patients, and SPECT medial temporal and posterior cingulate deficits are significant only in some analyses. Selective Reminding Test delayed recall, odor identification deficits, hippocampal volume, and entorhinal volume are the variables identified for the initial algorithm to predict MCI conversion to AD. This head-to-head comparison of predictors, and the initial predictive algorithm for MCI conversion to AD, are novel and unique, and have potential clinical impact. Completion of 9-year follow-up is needed to accurately identify all MCI converters to AD, and to improve accuracy for evaluation of the hypothesized predictors and the predictive algorithm. In order to achieve eventual clinical application, it is critical to build upon the ongoing study's findings by enhancing the signal to noise ratio, improving predictive accuracy by modifying and expanding specific assessments, and independently validating the findings in a new sample. Therefore, in a new study, 140 patients with amnestic MCI (with or without other cognitive domain deficits) and 50 control subjects will be recruited and systematically followed to test new potential predictors of MCI conversion to AD. These include more sensitive and specific cutting-edge brain imaging techniques that should enhance signal to noise ratio and predictive accuracy: improved structural MRI, [18F]-FDG, and p-amyloid imaging with [11C]-6-OH-BTA-1 (Pittsburgh compound B). New [11C]-6-OH-BTA-1 PET pilot data are very promising. Informant report of cognitive decline will be added, and the functional, olfactory and neuropsychological predictors will be modified and improved. In subjects who agree to lumbar puncture, CSF tau, phosphorylated tau (P- tauisi), AB42 and F2-isoprostanes will be assessed at baseline. The primary analyses in the new study will be restricted to the new study sample. Before combining the amnestic MCI patients across the two samples to generate the predictive algorithm for predictors common to the two studies, we will rediagnose the ongoing study sample's follow-up visits using the improved new study methods, ensure comparability in demographics/clinical features, and require that specific predictor by sample interaction criteria be met. Relevance. There are several clinical and public health implications. Accurate prediction of MCI conversion to AD will help identify patients who need early treatment, facilitate patient/family planning for their own future, and improve patient selection for clinical trials.